Automatic Diagnosis Method for Short Circuit Faults in Power Measurement Instruments Based on CNN

Abstract

The accuracy of short-circuit fault diagnosis methods for power measuring instruments is still low, which can directly lead to abnormal power data statistics. To address this issue, a convolutional neural network is used to construct an automatic diagnosis model for short circuit faults in power measuring instruments. The model utilizes wavelet packet energy spectrum to extract crucial features from the signal. The model identifies and determines the fault type based on the obtained operation characteristic data and the set short-circuit fault diagnosis criteria. The Sparrow optimization algorithm is used to optimize the model's parameters and enhance its performance. Experimental analysis revealed that after the signal features were extracted by using the wavelet energy spectrum distribution, the error value fluctuated in the range of 0.00 ~ 0.04 with regard to the measured value and was mainly around 0.02. The extracted features achieved high accuracy levels. The designed model exhibited an average diagnostic accuracy of 97.03%, surpassing the other three models by 9.24%, 7.10%, and 4.25%. The presented model can improve the precision and productivity of fault detection, support the safety and reliability of power system operations, and facilitate the collection and analysis of power usage
data.